A few days ago in a post on Greg Mankiw we got into a discussion of stock market valuation where vtcouger wondered about using 10 year trailing earnings to evaluate stocks. vtcouger wondered why Benjamin Graham recommended using 10 year trailing earnings rather then one year trailing earnings. Remember, Benjamin Graham was writing in the 1930s when earnings volatility was much greater then since WW II. Moreover, earnings growth was much weaker
As the chart shows, from 19871 to 1940 EPS growth was only 2% and the volatility was much greater. It was so great that if you used trailing one year earning growth to estimate stock market valuation it could generate major problems. For example, the day FDR was elected president was one of the best time ever to buy stocks. But based on trailing one year earnings the market PE was 25,which said the market was massively overvalued. To get around that problem Graham and Dodd advocated using trailing 10 year earnings. This measure of EPS would include both peak earning and trough earnings and so generate an estimate of normalized or trend earnings so that investors would not be mislead by using peak or trough earnings to calculate stock valuations.

Just as an aside, note that from 1875 to 1900 EPS did not grow. But we see many people talking about that as a great era of prosperity and deflation. I wonder how many of those who thing the late 1800s was so great would really be happy with decades of falling earnings or profits.

Since WW II earnings growth has become much more stable and market valuation has come to depend on future earnings growth rather than dividends.
Corporation now retain a much larger share of earnings and investors buy stocks much more on the basis of expected future growth from these retained earnings rather than current dividends that were so important to Graham and Dodd. Note, when they wrote in the 1930s expected earnings growth was 2% and now it is 7% and in the late 1990s bubble it was even higher.

But investors still have the problem that basing valuation of peak earnings or trough earnings can distort valuations even though EPS growth is now much more stable then it was before WW II. For the market as a whole you can use the Graham and Dodd practice of using trailing 10 year EPS that includes both peak and trough EPS. Another way to do it is to use trend EPS growth. If you look at the above chart on EPS since 1960 you see the dotted 7% EPS growth trend. So rather than using actual trailing EPS, or forecast EPS, use this trend line to estimate market valuation. This chart does just that.

In addition I have shown my estimate of what the PE should be given current interest rates. The estimate is based on the actual relationship between 1960 and 1996 before the late 1990s bubble so it excludes the bubble from the regression results.

I hope this explains things to your satisfaction vtcouger.

Note: my EPS data is reported EPS until 1989 and since 1989 it is operating EPS.

Note: read Brad Delong at